- Email: [email protected]
Is human papillomavirus testing an effective triage method for detection of high-grade (grade 2 or 3) cervical intraepithelial neoplasia?
Is human papillomavirus testing an effective triage method for detection of high-grade (grade 2 or 3) cervical intraepithelial neoplasia? Ervin Adam, MD,a,b Raymond H. Kaufman, MD,b Zuzana Berkova, PhD,a Joseph Icenogle, PhD,c and William C. Reeves, MDc Houston, Texas, and Atlanta, Georgia OBJECTIVE: Our purpose was to assess the usefulness of the polymerase chain reaction assay for detection of human papillomavirus infection for prognostic value in the triage strategies for high-grade (grade 2 or 3) cervical intraepithelial neoplasia in women referred for colposcopy after abnormal Papanicolaou smears. STUDY DESIGN: A total of 1007 women referred to a colposcopic clinic providing care for an indigent population were studied. Four hundred fifty-four women were referred after two Papanicolaou smears reported as atypical squamous cells of undetermined significance or low grade-squamous cervical intraepithelial lesion, and 553 were referred after a single smear reported as high-grade squamous intraepithelial lesion. All women had a cervical smear, colposcopy-directed biopsy, and endocervical curettage performed. A sample for human papillomavirus deoxyribonucleic acid detection by polymerase chain reaction was obtained. RESULTS: High-risk human papillomavirus types were detected in 463 (46%) of 1007 women studied. There was a significant increase of the frequency of high-risk human papillomavirus by the increasing severity of biopsy findings ranging from 32.7% in women without cervical intraepithelial neoplasia on biopsy to 60% in women having grade 2 or 3 on the biopsy specimen. Women having a negative Papanicolaou smear found to have high-risk human papillomavirus deoxyribonucleic acid at the time of colposcopy had a significantly higher rate of grade 2 or 3 cervical intraepithelial neoplasia on the biopsy specimen than did women without high-risk human papillomavirus. There was no such difference observed in women with a cytologic finding of low- or high-grade squamous intraepithelial lesions at the time of colposcopy. The polymerase chain reaction assay appears to be more sensitive than the commercial human papillomavirus profile test. The positive predictive value for grade 2 or 3 cervical intraepithelial neoplasia of both tests was similar (21.7% and 22.8%, respectively). CONCLUSION: The human papillomavirus is associated with high-grade cervical intraepithelial neoplasia, but the screening for human papillomavirus deoxyribonucleic acid does not have prognostic value in women reported as having atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions on two precolposcopy Papanicolaou smears. (Am J Obstet Gynecol 1998;178:1235-44.)
Key words: Human papillomavirus, polymerase chain reaction, cervical intraepithelial neoplasia, screening
The decreasing incidence of cervical cancer during the last four decades is attributed to widespread cytologic screening. Epidemiologic observations have suggested that sexually transmitted agents may be involved in the development of carcinoma of the cervix. On the basis of numerous epidemiologic and experimental studies, the
From the Division of Molecular Virologya and the Department of Obstetrics and Gynecology,b Baylor College of Medicine, and the Centers for Disease Control and Prevention.c Supported by contract No. 200-92-0537, Centers for Disease Control and Prevention, Atlanta, Georgia. Certificate of Merit Award, presented at the Sixty-fifth Annual Meeting of The Central Association of Obstetricians and Gynecologists, Scottsdale, Arizona, October 29–November 1, 1997. Reprint requests: Raymond H. Kaufman, MD, Baylor College of Medicine, Department of Obstetrics and Gynecology, 6550 Fannin, Suite 701, Houston, TX 77030. 6/6/89840
human papillomavirus (HPV) is considered the most plausible pathogenic agent, and type 16 is implicated as the highest risk factor in the etiology of cervical carcinogenesis.1 Consequently, for diagnostic and prognostic purposes, different test procedures have been developed for detection of HPV deoxyribonucleic acid (DNA) in cervical cells to serve as markers for neoplasia. Such assays claiming gradually higher sensitivity have become commercially available. Cervical cancer is a worldwide problem, and development of screening strategies should consider their applicability in different populations of various social and cultural backgrounds and with different standards and quality of health care. Therefore we initiated a longitudinal prospective study in a population that uses predominantly public health–supported services. One of the goals is to assess the usefulness of detecting HPV infec1235
1236 Adam et al.
June 1998 Am J Obstet Gynecol
tion for prognostic value in screening strategies. We report here data from a study using the polymerase chain reaction (PCR) technique for detection of HPV DNA as a triage for cervical intraepithelial neoplasia grades 2 or 3 and discuss the utility of various testing protocols.2
Texas. Only women not treated for cervical abnormalities during the last 2 years preceding referral were enrolled in the study. At the time of enrollment, information related to demographic, reproductive, and selected behavior factors were collected. From the original cohort of 1128 women, 1023 who had all examinations completed were available for this study. Examination of patients. At the clinic visit after referral a Papanicolaou smear with a Cytobrush endocervical brush (Medscand) and spatula was obtained, and the dacron swab provided with the HPV profile kit was used for collecting an endocervical sample for HPV testing. This was also used to provide a sample for PCR. In addition, a cervical vaginal lavage with 10 ml of normal saline solution was obtained. Colposcopy examination indicated any abnormal areas for cervical biopsies, which were taken after an endocervical curettage. The colposcopy findings and sites of biopsies were recorded on a standardized form. A blood sample was obtained from women who agreed to blood drawing (not an enrollment condition). Diagnostic tests. The enrollment Papanicolaou smear was classified according to the Bethesda classification system in the same laboratory by a single individual. The same person evaluated all biopsy specimens (R.H.K.), which were reported as HPV changes; cervical intraepithelial neoplasia grade 1, 2, or 3; carcinoma in situ; invasive squamous cell carcinoma; adenodysplasia; adenocarcinoma in situ; or adenocarcinoma. The PCR technique was used for HPV screening. HPV DNA was amplified by a general primer pair within the L1 open-reading frame. The primers MY09/MY 11 were
Material and methods The population studied has been previously described.2 Briefly, patients with a Papanicolaou smear reported as high-grade squamous intraepithelial lesion or two smears with atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesion were referred from city and county clinics caring for the indigent population in Harris County, Texas, to the colposcopy clinic in the Department of Obstetrics and Gynecology of Baylor College of Medicine in Houston, Table I. Type-specific primers Type-specific primers 6.1 6.2 11.1 11.2 16.1 16.2 18.1 18.2 31.1 31.2 33.1 33.2 35.1 35.2
Primer sequence
Reference
+5´ TAGTGGGCCTATGGCTCGTC 3´ –5´ TCCATTAGCCTCCACGGGTG 3´ +5´ GGAATACATGCGCCATGTGG 3´ –5´ CGAGCAGACGTCCGTCCTCG 3´ +5´ TGCTAGTGCTTATGCAGCAA 3´ –5´ ATTTACTGCAACATTGGTAC 3´ +5´ AAGGATGCTGCACCGGCTGA 3´ –5´ CACGCACACGCTTGGCAGGT 3´ +5´ ATGGTGATGTACACAACACC 3´ –5´ GTAGTTGCAGGACAACTGAC 3´ +5´ ATGATAGATGATGTAACGCC 3´ –5´ AGTATTTTCTCCTGCACTGC 3´ +5´ CAGAAGTGGACAGACATTGT 3´ –5´ ATGCATACTCCATATGGCTG 3´
5 5 5 5 5 6 7
Table II. HPV types related to histologic findings HPV types
Histologic finding No CIN CIN 1/HPV CIN 2 CIN 3/ CIS Invasive carcinoma# All χ2 for trend Statistical significance
No. 269 477 134 123 4 1007
16* 66 (24.5) 176 (36.9)ll 67 (50.0) 64 (52.0)¶ 373 (37.0) 36.35 p < 0.001
18† 12 (4.5) 22 (4.6) 4 (3.0) 5 (4.1) 1 (25.0) 44 (4.4)
31, 33, 35‡ 10 (3.7) 22 (4.6) 9 (6.7) 5 (4.1) 46 (4.6)
Any high-risk type
All multiple high-risk types
88 (32.7) 220 (46.1) 80 (59.7) 74 (60.2) 1 (25.0) 463 (46.0) 35.98 p < 0.001
38 (14.1) 103 (21.6)ll 28 (20.9) 26 (21.1) 195 (19.4)
Number in parentheses is percent. OR, Odds ratio; CI, confidence interval; CIN, cervical intraepithelial neoplasia; CIS, carcinoma in situ. *Includes type 16 and its combinations with other types. †Includes type 18 and its combinations with types 31, 33, ‡Includes type 31 or 33 or 35. §Odds ratio for any high-risk type. llIncludes one case of adenodysplasia. ¶Includes one case of adenocarcinoma in situ. #Not included in trend analysis. **Includes one case of adenocarcinoma.
and 35.
Adam et al. 1237
Volume 178, Number 6 Am J Obstet Gynecol
used to identify known HPV types as 6, 11, 16, 18, 30, 31, 33, 35, 39, 40, 42, 43, 45, 51, 52, 53, 54, 55, 57, 58, and 59 and 20 unidentified types.3, 4 All samples with negative results with the consensus primers were tested for β-globin with primers GH20 and PC04 for amplification.4 The samples positive for HPV DNA were then assayed with type-specific primers for types 16, 18, 31, 33, and 35. Samples negative for these high-risk types were then tested for low-risk types with primers for types 6 and 11 (Table I).5-7 The PCR conditions used for consensus primers were preheating to 94° C for 5 minutes, and a total of 35 PCR amplification cycles were performed with use of thermocycle-step parameters of 94° C for 30 seconds, 55° C for 30 seconds, and 72° C for 1 minute. The conditions used for the type-specific primers were the same except that 40 amplification cycles were applied. The amplification products were electrophoresed in a 3% agarose gel and were visualized by ethidium bromide. Both HPV testing protocols were performed blinded. Statistical methods. The differences in proportions were calculated by Yates’ corrected χ2 statistics or twotailed Fisher’s exact tests when indicated. Cornfield’s confidence limits were obtained for the odds ratio or exact limits if indicated.8 Positive predictive value, sensitivity, and specificity of the HPV test were calculated for cervical intraepithelial neoplasia grades 2 or 3 in the biopsy specimen.9 Results Three samples were not available for PCR testing, and 13 samples were negative for β-globin. Thus 1007 women were included in this study.
The distribution of high-risk types of HPV detected by PCR was similar in all races. The relationship of different HPV types identified by PCR to histologic findings on enrollment is illustrated in Table II. HPV type 16 alone or in combination with other high-risk types was found in 373 (37.0%) of all women examined or in 56.9% of 655 women with any HPV DNA present in the specimen tested. HPV 16 was identified in 24.5% of women without significant biopsy findings, gradually increasing by severity of biopsy grade to 52.0% in women with cervical intraepithelial neoplasia grade 2 or 3. This trend was highly significant (p < 0.001). HPV 18 was detected in 4.4% and types 31, 33, and 35 in 4.6% of women, respectively, representing 13.7% of all HPVpositive cases. There were 463 (46.0%) samples with at least one high-risk type identified. A total of 70.7% of all samples were positive for HPV by consensus primers. Of these 463 high-risk HPV-positive women, 80.6% had HPV 16. Multiple high-risk types were detected in 19.4% of all women and in 29.8% of samples positive for HPV. Only low-grade HPV types were found in 11.3% of all samples or in 17.4% of the 655 HPV positives. Neither the low-risk nor the unidentified types showed a significant trend with histologic grade. Overall, the odds ratio for any high-risk type gradually increased to 3.11 for cervical intraepithelial neoplasia grade 2 or 3 with a significant trend (p < 0.001). There were 454 women referred to the clinic with smears reported as atypical squamous cells of undetermined significance (n = 35) or low-grade squamous intraepithelial lesions (n = 419) (Table III.) The clinic (follow-up) Papanicolaou smear was reported as negative in
HPV types
6 or 11 27 (10.0) 60 (12.6) 8 (6.0) 18 (14.6)¶ 1 (25.0) 114 (11.3)
Unidentified type 25 (9.3) 33 (6.9) 7 (5.2) 12 (9.8) 1 (25.0) 78 (7.7)
Negative consensus primer 129 (48.0) 164 (33.4) 39 (29.1) 19 (15.4) 1 (25.0)** 352 (35.0) 41.06 p < 0.001
OR§
95% CI
Statistical significance
1.0 1.76 3.05 3.11 0.69
1.27-2.24 1.94-4.79 1.95-4.95 0.01-8.69
p < 0.001 p < 0.001 p < 0.001 p = 1.0
1238 Adam et al.
June 1998 Am J Obstet Gynecol
Table III. Follow-up Papanicolaou smear, high-risk HPV by PCR, and biopsy in 454 women with atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions in referral Papanicolaou smear (N = 454) Follow-up Papanicolaou smear No SIL (n = 266, 58.6%) High-risk HPV–
High-risk HPV+
Biopsy 184 (69.2)* 82 (30.8) No CIN 75 (40.8)† 26 (31.7) CIN 1/HPV 100 (54.3) 40 (48.8)‡ CIN 2/3 9 (4.9)§ 16 (19.5)§ OR for CIN 2/3 4.71 95% CI 1.85-12.22 Statistical significance p < 0.001 Positive predictive value 19.5 Sensitivity 64.0 Specificity 72.6
ASCUS/LSIL (n = 154, 33.9%) High-risk HPV–
HSIL (n = 34, 7.5%)
High-risk HPV+
High-risk HPV–
High-risk HPV+
91 (59.1) 16 (17.6) 60 (65.9) 15 (16.5) 1.18 0.45-3.19 p = 0.88 16.5 62.5 41.5
9 (26.5) 0 4 (44.4) 5 (55.6)
25 (73.5) 2 (8.0) 11 (44.0) 12 (48.0)ll
63 (40.9) 18 (28.6) 36 (57.1) 9 (14.3)
1.00 0.12-4.41 p = 0.74 48.0 70.6 23.5
All (N = 454, 100.0%) High-risk HPV–
High-risk HPV+
256 (56.4) 198 (43.6) 93 (36.3) 44 (22.2) 140 (54.7) 111 (56.1) 23 (9.0) 43 (21.7) 2.81 1.58-5.03 p < 0.001 21.7 65.2 60.1
SIL, Squamous intraepithelial lesions; ASCUS, atypical squamous cells of undetermined significance; LSIL, low-grade squamous intraepithelial lesions; HSIL, high-grade squamous intraepithelial lesions; CIN, carcinoma in situ; OR, odds ratio; CI, confidence interval. *Number in parentheses is percent of column. †Number in parentheses is percent of Papanicolaou diagnoses. ‡Includes one case of adenodysplasia. §Includes one case of adenocarcinoma in situ. llIncludes one case of follow-up Papanicolaou smear showing invasive carcinoma.
58.6% of cases, remained unchanged in 33.9%, and showed a high-grade squamous intraepithelial lesion in 7.5% of women. Of the 266 women who had a negative clinic smear, 25 (9.4%) had biopsy-proved cervical intraepithelial neoplasia grade 2 or 3. In this group, 19.5% of women with high-risk HPV were found to have grade 2 or 3 on biopsy in contrast to only 4.9% of women without high-risk HPV. In 154 women with atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions in the clinic smear, the finding of highgrade cervical intraepithelial neoplasia in biopsy specimen was not significantly different in the high-risk HPV positive and negative patients (14.3% and 16.5%). The same was true for 34 women whose smear was reported as high-grade lesions (55.6% to 48%). A significantly higher odds ratio for cervical intraepithelial neoplasia grade 2 or 3 (4.71%) was observed in women with a negative clinic smear with a positive predictive value of 19.5%, a sensitivity of 64%, and a specificity of 72.6%. No such significant finding can be seen in women with atypical squamous cells of undetermined significance or lowor high-grade squamous intraepithelial lesions in the clinic smear. The sensitivity of the test exceeded 60%, but the specificity dropped substantially to 41.5% and 23.5% in the latter two groups. Five hundred fifty-three women were referred with a Papanicolaou smear reported as high-grade squamous intraepithelial lesions (Table IV). In this group of women
the clinic cytologic findings were reported as normal in 41.0%, atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions in 32.5%, and high-grade squamous intraepithelial lesions in 26.4%. In the 227 women found to have normal clinic cytologic findings, cervical intraepithelial neoplasia grade 2 or 3 was found in a significantly higher proportion (33.7%) in women with high-risk HPV than in women without high-risk HPV (13.2%). There was no such significant difference observed in the other two groups. The most disputed triage approach is in women with atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions reported in their Papanicolaou smears. We compared the results of PCR testing with findings of our previous study with use of the HPV profile assay in this group of women (Table V).2 The HPV profile detected high-risk HPV in 162 (35.7%) specimens, and the PCR was positive for highrisk HPV in 198 women (43.4%). In addition, the PCR detected HPV with consensus primers in 31 women (7%) not identified by the limited number of type-specific primers used in PCR in this study. Both tests had a similar positive predictive value (22.8% and 21.7%), but the PCR had a somewhat higher sensitivity (65.2% compared with 56.1%) but lower specificity (60.1% compared with 67.8%) than the HPV profile in patients found to have cervical intraepithelial neoplasia grade 2 or 3 on the biopsy specimen. The Papanicolaou smear taken at the
Adam et al. 1239
Volume 178, Number 6 Am J Obstet Gynecol
Table IV. Follow-up Papanicolaou smear, high-risk HPV by PCR, and biopsy in 553 women with high-grade squamous intraepithelial lesions in referral Papanicolaou smear (N = 553) Follow-up Papanicolaou smear No SIL (n = 227, 41.0%) High-risk HPV–
ASCUS/LSIL (n = 180, 32.5%)
High-risk HPV+
High-risk HPV–
Biopsy 141 (62.1)* 86 (37.9) No CIN 63 (44.7)† 19 (22.1) CIN 1/HPV 59 (41.8) 38 (44.2) CIN 2/3 19 (13.5)† 29 (33.7) OR for CIN 2/3 3.27 95% CI 1.61-6.66 Statistical significance p < 0.001 Positive predictive value 33.7 Sensitivity 60.4 Specificity 68.2
89 (49.4) 13 (14.6) 44 (49.4) 32 (36.0)
HSIL (n = 146, 26.4%)
High-risk HPV+
91 (50.6) 11 (12.1) 54 (59.3) 26 (28.6) 0.71 0.36-1.40 p = 0.37 28.6 44.8 46.7
High-risk HPV–
High-risk HPV+
58 (39.7) 88 (60.3) 12 (20.7) 14 (15.9) 14 (24.1) 17 (19.3) 32 (55.2) 57 (64.8) 1.49 0.72-3.11 p = 0.32 64.8 64.0 45.6
All (N = 553, 100.0%) High-risk HPV–
High-risk HPV+
288 (52.1) 265 (47.9) 88 (30.6) 44 (16.6) 117 (40.6) 109 (41.1) 83 (28.8) 112 (42.3) 1.81 1.25-2.61 p < 0.005 42.3 57.4 57.3
SIL, Squamous intraepithelial lesions; ASCUS, atypical squamous cells of undetermined significance; HSIL, high-grade squamous intraepithelial lesions; CIN, cervical intraepithelial neoplasia; OR, odds ratio; CI, confidence interval. *Number in parentheses is percent of column. †Number in parentheses is percent of Papanicolaou diagnoses. ‡Includes one case of adenocarcinoma.
Table V. Histologic findings and high-risk HPV in 454 women referred with atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions on cytologic study Histologic findings High-risk HPV
No.
No CIN (n = 137, 30.2%)
HPV profile
Positive Negative
162 292
39 (24.1) 98 (33.6)
86 (53.1) 165 (56.5)*
37 (22.8) 29 (9.9)†
PCR
Positive Negative
198 256
44 (22.2) 93 (36.3)
111 (56.1)* 140 (54.7)
43 (21.7)‡ 23 (9.0)‡
Cytologic findings (clinic Papanicolaou smear)
ASCUS or worse Negative
188
36 (19.1)
111 (59.0)
41 (21.8)
266
101 (38.0)
140 (52.6)*
25 (9.4)†
HPV assay
CIN 1/HPV (n = 251, 55.3%)
CIN 2/3 (n = 66, 14.5%)
OR for CIN 2/3, 95% CI, statistical significance 2.68, 1.53-4.72, p < 0.001 2.81, 1.58-5.03, p < 0.001 2.69, 1.52-4.77, p < 0.001
Screening properties PPV 22.8, SE 56.1, SP 67.8 PPV 21.7, SE, 65.2, SP 60.1 PPV 21.8, SE 62.1, SP 62.1
Number in parentheses is percent of row. CIN, Cervical intraepithelial neoplasia; OR, odds ratio; CI, confidence interval; PPV, positive predictive value; SE, sensitivity; SP, specificity; ASCUS, atypical squamous cells of undetermined significance. *Includes one case of adenodysplasia. †Includes one case of adenocarcinoma in situ. ‡Includes two cases of adenocarcinoma in situ.
time of colposcopy had very similar screening properties with a positive predictive value of 21.8%, a sensitivity of 62.1%, and a specificity of 62.1%. Comment In our previous study of the same population2 we concluded that the approved commercial HPV profile test kit is not sensitive enough for the routine screening of women with atypical Papanicolaou smears for the purpose of triaging patients for immediate colposcopy. Therefore we used the PCR assay, considered the most
sensitive method for HPV DNA detection. The most frequent types of high or intermediate risk (types 16, 18, 31, 33, and 35) and the two most common low-risk HPV types (types 6 and 11) were studied. Our main interest was in finding the prevalence of high-risk types; therefore testing for low-risk types was done only in samples negative for high-risk types. Thus multiple types, when detected, do not include types 6 and 11. When we associate the HPV findings with biopsy findings, it is apparent that 57% of all HPV-positive samples were represented by type 16 and an additional 13.7%
1240 Adam et al.
were of other high-risk types. HPV 16 was identified in 80.6% of patients who had any high-risk types detected. The increasing trend of prevalence of type 16 by higher grades of severity of cervical intraepithelial neoplasia was significant. Women referred to the colposcopy clinic with atypical squamous cells of undetermined significance or lowgrade squamous intraepithelial lesions were seen by different physicians, and the referral cytology examinations were performed in different laboratories. The clinic Papanicolaou smears and biopsy specimens taken at the time of colposcopy were processed in the same laboratory and evaluated by the same pathologist (R.H.K.) without knowing the HPV status of the patient. Thus, if there were any bias in the diagnosis, it would occur with consistency. Women referred with a smear reported as atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions had a negative clinic Papanicolaou smear in 58.6% of cases. It is possible that the referral Papanicolaou smear was in many cases overdiagnosed; on the other hand, the clinic smear might have had, in some cases, false-negative results because of sampling differences. Nevertheless, in this group of patients the incidence of biopsy-proved cervical intraepithelial neoplasia grades 2 and 3 was relatively low—9.4% compared with that in women whose repeat Papanicolaou smears were reported as atypical squamous cells of undetermined significance or low- (15.6%) or high-grade squamous intraepithelial lesions (50%). Similar findings were observed in women referred for high-grade squamous intraepithelial lesions in their Papanicolaou smears. The clinic smear was within the normal range in 41% of all the patients studied. In 21.1% of these women cervical intraepithelial neoplasia grade 2 or 3 was found in the biopsy specimen. Women with atypical squamous cells of undetermined significance or lowgrade squamous intraepithelial lesions in the clinic Papanicolaou smear had cervical intraepithelial neoplasia grade 2 or 3 in 32.2% of their biopsy specimens. A total of 61% of women whose clinic smear was reported as high-grade lesions were found to have cervical intraepithelial neoplasia grade 2 or 3 on the biopsy specimen. The question arises as to how HPV testing would contribute to the detection of cervical intraepithelial neoplasia grades 2 or 3. Women with normal cytologic findings in the clinic smear, whether they were referred with lowor high-grade squamous intraepithelial lesions and detectable high-risk HPV DNA, had a significantly higher rate of biopsy-proved grades 2 and 3 neoplasia than did women without HPV infection. In contrast, women with any cytologic abnormality in the clinic smear had a frequency of cervical intraepithelial neoplasia grade 2 or 3 in the biopsy specimen irrespective of their HPV status. Thus the HPV testing appears helpful for the proper diagnosis and treatment in women with a normal follow-up
June 1998 Am J Obstet Gynecol
Papanicolaou smear after a previous abnormal smear was found. In spite of this, a significant number of patients with a negative HPV test and a negative follow-up smear were found to have grade 2 or 3 neoplasia on the biopsy specimen: 9 of 25 (36%) women referred with atypical squamous cells of undetermined significance or lowgrade squamous intraepithelial lesions and 19 of 48 (39.6%) referred with high-grade squamous intraepithelial lesions. Our results indicate that a repeat Papanicolaou smear showing a similar or more severe abnormality than the first one, irrespective of the HPV testing, is an indication for colposcopy with directed biopsies. A clinic smear at the time of colposcopy does not appear to contribute to further management, as previously noted by Young et al.10 Evidently, the sensitivity and specificity of the HPV assay are of great importance. The comparison of the two HPV assays used in our study also shows a difference, with the PCR being more sensitive in general. The PCR in our study detected HPV in 154 of 257 patients with cervical intraepithelial neoplasia grade 2 or 3. In addition, 19 women (7.3%) were found to have unidentified HPV types. Cox et al.,11 on using the Hybrid Capture test, reported 14 (93.3%) high-risk HPV positives in 15 women with grade 2 or 3 neoplasia on the biopsy specimen. Schneider et al.,12 with use of the same assay, found only 50% (19/38) high-risk HPV in biopsy-proved grade 2 or 3 neoplasia. Wright et al.,13 also using the Hybrid Capture test, found high-risk HPV in 6 of 11 (55%) biopsy-proved grade 2 or 3 neoplasia in women referred because of a smear reported as atypical squamous cells of undetermined significance and in 33 of 39 (84.6%) with grade 2 or 3 neoplasia on the biopsy specimen in women referred for low-grade squamous intraepithelial lesions. Sun et al.14 reported high-risk HPV in 79% of 105 cases with grade 2 or 3 neoplasia or cancer. With use of a PCR assay Cuzick et al.15 identified high-risk HPV types in 58% of women with grade 2 neoplasia and in 84% of women with grade 3 neoplasia on histologic examination. There is a considerable variation in the sensitivities for HPV testing even when a single test is used, possibly because of patient selection or study design. There is no confirmed evidence yet that HPV causes all invasive cervical cancers, although such a possibility cannot be excluded.16 Recently Shah et al.17 did not find differences in the presence of high-risk HPV in patients with cervical intraepithelial neoplasia grade 3 on the biopsy specimen that did or did not progress to invasive cancer. Kiviat and Koutsky18 and Koutsky et al.19 concluded that cervical intraepithelial neoplasia is an early manifestation of HPV infection. They also found, after adjustment for HPV, an association of chlamydia and cytomegalovirus infection with development of cervical neoplasia. Downey et al.20 reported that development of cervical intraepithelial neoplasia grade 2 or 3 is higher in
Volume 178, Number 6 Am J Obstet Gynecol
women with biopsy-proved grade 1 neoplasia at study entry than in women with normal histologic features. There was no difference in disease-free probability in relation to HPV-16 infection. Wheeler et al.21 reported on the limitation of a single-point detection of HPV DNA for assessment of a woman’s HPV status. All these observations definitely indicate a major but not exclusive role of HPV infection in cervical neoplasia. Assays identifying a higher proportion of high-risk HPV in cervical intraepithelial neoplasia grade 2 or 3 also detect an increased prevalence of this infection in grade 1 neoplasia and even in women without significant pathologic findings. To test all patients with abnormal Papanicolaou smears for high-risk HPV has its advantages and substantial drawbacks. As mentioned, a positive test would support the necessity of a colposcopic examination and enhance the detection of a grade 2 or 3 neoplasia on the biopsy specimen of a woman referred for a cytologic abnormality who has a follow-up smear reported as normal. On the other hand, solely relying on these tests would still miss a considerable number of patients with grade 2 or 3 neoplasia. So far, there is no good evidence to support the cost effectiveness of HPV testing in clinical practice.2 The majority of cases of cervical intraepithelial neoplasia grade 1 regress, including many cases in our experience with detected high-risk HPV. A positive test indicates a higher probability of a diagnosis of grade 2 or 3 neoplasia, but a negative test does not exclude it. Thus the use of these tests at present should remain in the research domain.
REFERENCES
1. Zur Hausen H. Human papillomaviruses in the pathogenesis of anogenital cancer. Virology 1991;184:9-13. 2. Kaufman RH, Adam E, Icenogle J. Relevance of human papillomavirus screening in management of cervical intraepithelial neoplasia. Am J Obstet Gynecol 1997;176:87-92. 3. Manos MM, Ting Y, Wright DK, Lewis AJ, Broker TR, Wolinsky SM. Cancer cells 7: molecular diagnostics of human cancer. Cold Spring Harbor: Cold Spring Harbor Laboratory; 1989. 4. Bauer HM, Ting Y, Greer CE, Chambers JC, Tashiro CJ, Chimera J, et al. Genital human papillomavirus infection in female university students as determined by a PCR-based method. JAMA 1991;265:472-7. 5. Van der Brule AJC, Meijer CLM, Bakels V, Kencmans P, Walboomers JMM. Rapid selection of human papillomavirus in cervical scrapes by combined general primer-mediated and type-specific polymerase chain reaction. J Clin Microbiol 1990;28:2739-43. 6. Rymark P, Forslund O, Hansson BG, Lindholm K. Genital HPV infection not a local but a regional infection: experience from a female teenage group. J Clin Microbiol 1990;28:739-43. 7. Forslund O, Hansson BG, Rymark P, Bjerrf B. Human papillomavirus DNA in urine samples compared with that in simultaneously collected urethra and cervix samples. Clin Microbiol 1993;31:1975-9. 8. Fless JL. Statistical methods for rates and proportions. New York: John Wiley; 1981. 9. Brookmyer R, Gail MH. AIDS epidemiology: a quantitative approach. New York: Oxford University Press; 1994.
Adam et al. 1241
10. Young NA, Naryshkin S, Bowman RL. Value of repeat cervical smear at the time of colposcopic biopsy. Diagn Cytopathol 1993;9:646-9. 11. Cox JT, Lorincz AT, Schiffman MH. Human papillomavirus testing by hybrid capture appears to be useful in triaging women with a cytologic diagnosis of atypical squamous cells of undetermined significance. Am J Obstet Gynecol 1995;172:946-54. 12. Schneider A, Zahn DM, Kirchmayr R. Screening for intraepithelial neoplasia grade 2/3: validity of cytologic study, cervicography, and human papilloma virus detection. Am J Obstet Gynecol 1996;174:1534-41. 13. Wright TC, Sun XW, Koulos J. Comparison of management algorithms for the evaluation of women with low-grade cytologic abnormalities. Obstet Gynecol 1995;85:202-10. 14. Sun X-W, Ferenczy A, Johnson D, Koulos JP, Lungu O, Richart RM, et al. Evaluation of the hybrid capture human papilloma virus deoxyribonucleic acid detection test. Am J Obstet Gynecol 1995;173:1432-37. 15. Cuzick J, Terry G, Ho L, Hollingworth T, Anderson M. Type-specific human papillomavirus DNA in abnormal smears as a predictor of high-grade cervical intraepithelial neoplasia. Br J Cancer 1994;69:167-71. 16. Zur Hausen H. Are human papillomavirus infections not necessary or sufficient causal factors for invasive cancer of the cervix? Int J Cancer 1995;63:315-6. 17. Shah KV, Kessis TD, Shah F, Gupta JW, Shibata D, Jones RW. Human papillomavirus investigation of patients with cervical intraepithelial neoplasia 3, some of whom progressed to invasive cancer. Int J Gynecol Pathol 1996;15:127-30. 18. Kiviat NB, Koutsky LA. Specific human papillomavirus types as the causal agents of most cervical intraepithelial neoplasia: implications for current views and treatment. J Natl Cancer Inst 1993;85:934-5. 19. Koutsky LA, Holmes KK, Critchlow CW, Stevens CE, Paavonen S, Beckman AM. A cohort study of the risk of cervical intraepithelial neoplasia grade II or III in relation to papillomavirus. N Engl J Med 1992;327:1272-8. 20. Downey GP, Bavin PJ, Deery ARS, Crow J, Griffiths PD, Emery VC, et al. Relation between human papillomavirus type 16 and potential for progression of minor-grade cervical disease. Lancet 1994;344:432-5. 21. Wheeler CM, Greer CE, Becker TM, Hunt WC, Anderson SM, Manos MM. Short-term fluctuations in the detection of cervical human papillomavirus DNA. Obstet Gynecol 1996;88:261-8.
Discussion DR. MICHAEL HOPKINS, Akron, Ohio. I believe this report will influence our attitude toward the connection of HPV to premalignant disease of the cervix. The authors tried to determine whether HPV screening would help identify high-risk groups and their article contains a wealth of data with tables that extensively detail their results. The methods have been described and are appropriate for their population. There is no long-term patient follow-up. Thus their data are valid only for the incidence of cervical intraepithelial neoplasia and HPV at the initial colposcopic visit. In their report the authors compared referral Papanicolaou smear with follow-up Papanicolaou smear, both smears with biopsy specimen results; both smears with HPV results, and, last, biopsy specimen results with HPV results. In their analysis of >1000 patients a number of interesting details emerged and along with this a number of questions. The authors’ evaluated referred patients with colposcopic-directed biopsies. A PCR analysis was then performed for HPV. Interestingly, among patients with no dysplasia one third had high-risk HPV types. Also, one
1242 Adam et al.
half these patients without dysplasia did not have any identifiable HPV. Among the patients with moderate-severe dysplasia or carcinoma in situ, 60% had high-risk HPV. Interestingly, 25% had no identifiable HPV. Thus it would appear that identifying high-risk subtypes does not predict high-grade squamous intraepithelial lesions. Another interesting finding was the results when a follow-up or new Papanicolaou smear was done at the visit for colposcopy. When comparing the referral smear and the follow-up smear, a large discrepancy exists between the referral smear and the in-house or follow-up cytologic findings. Almost 500 patients each are in the lowgrade or high-grade referral groups. In the low-grade referral group, 60% had negative cytologic findings on repeat Papanicolaou smear, and only one third of the time did the low-grade cytologic findings agree with the referral cytologic findings. In >500 patients with highgrade changes 40% had negative cytologic findings and 33% had low-grade changes. Thus only 25% were confirmed to have high-grade changes on cytologic study, whereas 75%, an exceedingly large number, had a difference in their cytologic findings. Such widespread variations certainly highlight the inexact science of cytology. This, combined with the broad distribution of highrisk HPV types across the groups, is good evidence that colposcopy with biopsy will continue to be the standard for evaluation of abnormal cytologic findings. I took the liberty of combining some tables from the authors’ article and present these to highlight what I believe is a continuing problem in our management of premalignant disease. When using the new follow-up cytologic results, the authors report a very high rate of negative findings, although approximately 15% of these patients with negative results have moderate or severe dysplasia on the biopsy specimen. Among those with atypical squamous cells of undetermined significance or low-grade changes, approximately 25% have high-grade histologic findings. This is a continuing management problem with cytology. Most authors report an incidence in the 10% to 20% range of significant premalignant finding with low-grade cytologic findings. This highlights the problems with clinical management: The Papanicolaou smear is a very inexact science at best; a third of patients with no cervical intraepithelial neoplasia findings have high-risk HPV; patients may be exposed to a new viral agent that is unknown to the examining physician. I ask the following questions: (1) Was there any prolonged delay in the time from the referral Papanicolaou smear to the Papanicolaou smear obtained at colposcopy in the clinic? Also, were any of the referral smears done in the antepartum period with the patient not being seen in colposcopy clinic until the postpartum period? (2) Why is there such a large difference between the referral cytologic findings and those obtained at colposcopy? (3) Do the authors continue to do Papanicolaou smears at colposcopy? (4) How can we explain such a high rate of high-risk HPV in patients without any cervical intraepithelial neoplasia and what may be considered a low rate of high-risk HPV (grade 2 or 3) in patients with cervical intraepithelial neoplasia grade 3 or carcinoma in situ?
June 1998 Am J Obstet Gynecol
In closing, this is a significant report that suggests that perhaps HPV is not the complete answer to the etiology of premalignant disease. DR. JOHN CALKINS, Kansas City, Kansas. The authors raise an interesting question in their article. Specifically, what role does HPV testing play in triage of patients for identification of significant intraepithelial lesions of the cervix that would warrant treatment? As noted by the authors, we have witnessed a significant decrease in the incidence of cervical cancer since the introduction of Papanicolaou smear screening in the 1950s. It is intuitive that identification of precancerous changes and treatment thereof has played a major role in that reduction. In spite of that impact, the Papanicolaou smear has been much maligned of late. Clinicians are being inundated by changes in the technology, such as Pap-Net and Thin-Prep. Recognizing that HPV is the most plausible pathogenic agent for cervical carcinogenesis, identification of high-risk types is an appealing alternative to aid in selection of patients for more significant evaluation. In the current article the authors report on their experience using PCR in a study population on whom they had previously reported the use of the HPV profile testing kit for aid in triaging patients for immediate colposcopy. In that study they concluded that the HPV profile testing kit was not sensitive enough for routine screening for purposes of immediate triage. Recognizing that PCR is a more sensitive method for HPV detection, they reassessed their study population with use of consensus primers meant to identify in excess of 40 subtypes of HPV from samples obtained during the screening phase of their study. They did identify a significantly higher odds ratio for high-grade intraepithelial neoplasia in women with atypical squamous cells of undetermined significance and low-grade squamous intraepithelial lesions on referral Papanicolaou smears if their clinic smears were negative and their HPV testing was positive (with a positive predictive value of 19.5%, a sensitivity of 64%, and a specificity of 72.6%). However, if their clinic Papanicolaou smear was abnormal, then no such significant finding could be identified. When PCR was contrasted with the HPV profile detection kit, PCR was indeed positive in a higher percentage of women, 43% versus 36%. Both tests had similar positive predictor values, but the PCR was found to be more sensitive but less specific. The authors concluded that a positive HPV test indicates a higher probability of highgrade cervical intraepithelial neoplasia but does not exclude such a condition, and thus the test still should be viewed as a research tool. I will focus my questions on three areas. The first area relates to patient selection and specifically the referral Papanicolaou smear. Since the advent of the Bethesda system, many of us who provide colposcopic services have been plagued by an increase in referrals, largely as a result of the designation atypical squamous cells of undetermined significance. To triage patients for colposcopy, we will commonly repeat cytologic studies for this finding and for low-grade squamous intraepithelial lesions and
Adam et al. 1243
Volume 178, Number 6 Am J Obstet Gynecol
further evaluate only those patients with persistent squamous intraepithelial changes. As an individual who works directly with county health departments in the Kansas City area, I can attest to their anxiety about “observation” of minor Papanicolaou smear abnormalities, especially because of patient compliance. Looking at the portion of the patients in this study with these findings, 15% of the patients were found to have a high-grade lesion, more than one third of which would have been missed by PCR screening. The referral criteria in the current study could be considered relatively conservative because they required either a high-grade Papanicolaou smear result or consecutive findings of atypical squamous cells of undetermined significance or low-grade lesions for study entry. Interestingly, in the study design all patients received a “clinic Papanicolaou smear” and in that group almost half were read as normal. Do the authors have an idea regarding the incidence of atypical squamous cells of undetermined significance and low-grade squamous intraepithelial lesions in their county clinic study population, especially in light of the high incidence of negative Papanicolaou smear findings during the screening phase of their study? Speaking from my own experience, I had developed what I thought was a reasonable triage mechanism on the basis of the incidence of these in my own population. Unfortunately, with the advent of managed care, we are now required to use multiple laboratories for cytologic testing, which makes it much more difficult to track percentages and consequently continue to use such triage mechanisms. My second question relates to study design and specifically “collection method.” The authors identified highrisk HPV types in roughly 60% of patients with highgrade disease on the biopsy specimen. Hatch et al.1 in a similar study published in 1995 using Hybrid Capture technology found a 74% incidence of high-risk HPV types with high-grade squamous intraepithelial lesions. Nuovo et al.2 a few years earlier quoted in excess of 90% positivity with high-grade lesions. Do the authors feel that the study design, specifically obtaining a Papanicolaou smear before HPV swab, may in any way affect the positivity of their detection rate? Additionally, do the authors feel that the timing of HPV collection is of significance? Wheeler et al.3 in 1996 published a report on short-term fluctuations in the detection of cervical HPV DNA with use of various methods. Specifically, with PCR they found a range between 21% and 47% monthto-month in their study population of 72 college-aged women, roughly one third of which had a history of an abnormal Papanicolaou smear. My third question relates to type-specific data identified in the article The authors identified type 16 in 57% of all HPV-positive samples and in excess of 80% of patients who had any high-risk type of HPV DNA detected. They noted an increasing trend of prevalence of type 16 by higher grades of severity of cervical intraepithelial neoplasia and concluded that it was significant. However, they felt that no such trend was observed with other HPV types. In light of the fact that the other high-risk HPV
types were identified alone in only single digit numbers of patients, I would question whether there is sufficient power to draw such conclusions. I would ask the authors if a power analysis was performed with regard to those conclusions. In conclusion, I applaud the authors in their attempt to improve the screening approach to patients with intraepithelial disease of the cervix and look forward to their follow-up data as it relates to screening protocols. Their conclusions support my personal opinion, that at this time HPV testing is a technique still in search of useful clinical application. REFERENCES
1. Hatch KD, Schneider A, Abdel-Nour MW. An evaluation of human papillomavirus testing for intermediate and high-risk types as triage before colposcopy. Am J Obstet Gynecol 1996;172:1150-7. 2. Nuovo GJ, Blanco JS, Leipzig S, Smith D. Human papillomavirus detection in cervical lesions nondiagnostic for cervical intraepithelial neoplasia: correlation with Papanicolaou smear, colposcopy, and occurrence of cervical intraepithelial neoplasia. Obstet Gynecol 1990;75:1006-11. 3. Wheeler CM, Greer CE, Becker TM, et al. Short-term fluctuations in the detection of cervical human papillomavirus DNA. Obstet Gynecol 1996;88:261.
DR. ROBERT J. CARPENTER, Houston, Texas. Would you or can you address the specific issue of those patients who on repeat Papanicolaou smears had negative results and what you think may have occurred in those that had been submitted to one of the forms of automated Papanicolaou smear review, whether you would have then seen a similar distribution to your biopsy data? DR. MELVIN V. GERBIE, Chicago, Illinois. How do you at this point counsel those 32% percent of patients who had negative biopsy specimens and yet still showed highrisk HPV changes? DR. KAUFMAN (Closing). The first comment was related to the time interval between the smears and how many were done ante partum or post partum. The time interval varied between 2 months and 1 year. The average was probably about 4 months between the time the last Papanicolaou smear was taken and the patients were seen in our clinic. There were very few of this population of individuals who were antepartum. I cannot give you the exact number, but I would say it’s probably <2% of the entire group. Why the difference between referral cytologic findings and the findings obtained in our clinic? There were several reasons. One could be a higher sensitivity, if you will, on the part of the pathologists working in the state laboratory. In general, the reporting that comes out of our state laboratory in Austin is extremely good. In fact, I did go back and review the majority of these referral smears; in many cases or in most cases there was general agreement, but in about 40% to 50% I tended to downgrade the reports. That may refer to the fact that I am undercalling or it could be the fact that they are overcalling, and that’s a question that I don’t think I can answer.
1244 Adam et al.
Also, another related factor is the technique of taking the smears. The smears were taken in our clinic by residents with faculty supervision, and the smears that went to the state laboratory were taken by a whole variety of individuals, family practitioners, obstetricians, nurse practitioners, etc. Do we still do Papanicolaou smears at the time of the colposcopy? In our clinic we do. There have been several studies to suggest that really is a waste of time. If you have a significantly abnormal smear and the patient is brought in for colposcopy, why bother to do another smear? It doesn’t really in the long run contribute to how you’re going to manage the patient. On the other hand, a point could be raised that if the colposcopic examination is normal, you don’t see anything abnormal, and you obtain tissue that does not show pathologic features, and the repeat Papanicolaou smear suggests high-grade disease, you better keep looking; whereas if it is reported as negative and your examination is negative and the biopsy specimens are negative, I think you can relax a bit and probably defer seeing the patient again for at least another 6 months. The question was raised as to high-risk HPV seen in a significant number of our patients with no pathologic features. I think this partly relates to the population of patients we are seeing. We are dealing with a very-highrisk population of individuals, and the prevalence of high-risk HPV in this population is quite high. Overall, in the group that we studied here, about 11% were found to have low-risk HPV; therefore 89% to 90% had a high-risk HPV type. In our population this is what we see. If you look at most studies that are done to examine the prevalence of HPV in a community, it may run as high as 40% or even higher in a given community, especially in a high-risk community such as we’re dealing with. The question was raised as to the prevalence of Papanicolaou smears reported as cervical intraepithelial neoplasia in our population. It again depends on which laboratory is doing the reporting. Somewhere between about 6% and 8% of the patients in our clinic population will have smears reported as atypical squamous cells of undetermined significance or higher. Did the method of collection have anything to do with it? Theoretically, with PCR it should not. It’s such a sensitive technique that even if you do a Papanicolaou smear first, you should expect to find plenty of cells with HPV DNA if, in fact, HPV DNA is present. The other point I think that relates to some of our findings is why we saw such a significant number of individuals with no high-risk HPV and yet grade 2 or 3 cervical intraepithelial neoplasia is the age of our population. About 30+%, a third of the patients, were >30 years old; our data, which have not yet been published, would sug-
June 1998 Am J Obstet Gynecol
gest that there was a distinct, significant decrease in the frequency of detection of HPV as women got older. In other words, in women >30 years old, the frequency of HPV detection was significantly less than it was in the group of patients between ages 15 and 25 years, for example, whereas the frequency of grade 2 and 3 cervical intraepithelial neoplasia increased significantly from the younger group to the older group. Therefore, we have a group of women >30 years old, many of whom were found to have grade 2 or 3 lesions and many of whom had negative HPV testing. We can discuss for another hour why that is. Is the HPV burned out, no longer present? Is it not detectable? I don’t think I can answer that at the present time, but those are just our findings. I don’t think we can draw any distinct conclusions about the other HPV types in relation to finding grade 2 or 3 cervical intraepithelial neoplasia. Certainly, when you lump all the high-risk HPV types, there’s a definite significant trend to find high-risk HPV in higher grades of cervical intraepithelial neoplasia. When it comes to specific types, besides type 16, I don’t think we had enough cases in the 18, 31, 33, and 35 group to come to any conclusions. Dr. Carpenter raised the point about whether automated cytologic studies would have changed some of our results; they may well have. In fact, we just completed a study sending a series of patients who had Papanicolaou smears reported as atypical squamous cells of undetermined significance in our laboratory for review; in 20% of the cases the smears were upgraded. All these patients, I might add, had had biopsies within a year’s time of the Papanicolaou smear being taken. There were 9 patients who were found to have grade 3 lesions, and in 6 of the 9 the smear was upgraded by automated cytologic study. As a matter of fact, right now we are in the midst of a study reviewing smears in our clinic reported as atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions. We tried to relate our findings to absence or presence of high-risk HPV. Dr. Gerbie asked about counseling in the patients who have a negative biopsy specimen and high-risk HPV. I just tell them they have this. The biopsy specimens are negative, and we’ll continue to follow them up. If the Papanicolaou smear stays negative, there’s no reason to be concerned, and as long as there’s nothing we can see in the biopsy, the likelihood of transmitting the infection is small. I don’t say it’s impossible because I don’t know; and we will follow them up very carefully at regular intervals, 6 months to 1 year, depending on what we found. If they have ever had an abnormal Papanicolaou smear, it will be 6-month intervals until we have three consecutive negative smears. Otherwise, we will see them at yearly intervals.
Key words: Human papillomavirus, polymerase chain reaction, cervical intraepithelial neoplasia, screening
The decreasing incidence of cervical cancer during the last four decades is attributed to widespread cytologic screening. Epidemiologic observations have suggested that sexually transmitted agents may be involved in the development of carcinoma of the cervix. On the basis of numerous epidemiologic and experimental studies, the
From the Division of Molecular Virologya and the Department of Obstetrics and Gynecology,b Baylor College of Medicine, and the Centers for Disease Control and Prevention.c Supported by contract No. 200-92-0537, Centers for Disease Control and Prevention, Atlanta, Georgia. Certificate of Merit Award, presented at the Sixty-fifth Annual Meeting of The Central Association of Obstetricians and Gynecologists, Scottsdale, Arizona, October 29–November 1, 1997. Reprint requests: Raymond H. Kaufman, MD, Baylor College of Medicine, Department of Obstetrics and Gynecology, 6550 Fannin, Suite 701, Houston, TX 77030. 6/6/89840
human papillomavirus (HPV) is considered the most plausible pathogenic agent, and type 16 is implicated as the highest risk factor in the etiology of cervical carcinogenesis.1 Consequently, for diagnostic and prognostic purposes, different test procedures have been developed for detection of HPV deoxyribonucleic acid (DNA) in cervical cells to serve as markers for neoplasia. Such assays claiming gradually higher sensitivity have become commercially available. Cervical cancer is a worldwide problem, and development of screening strategies should consider their applicability in different populations of various social and cultural backgrounds and with different standards and quality of health care. Therefore we initiated a longitudinal prospective study in a population that uses predominantly public health–supported services. One of the goals is to assess the usefulness of detecting HPV infec1235
1236 Adam et al.
June 1998 Am J Obstet Gynecol
tion for prognostic value in screening strategies. We report here data from a study using the polymerase chain reaction (PCR) technique for detection of HPV DNA as a triage for cervical intraepithelial neoplasia grades 2 or 3 and discuss the utility of various testing protocols.2
Texas. Only women not treated for cervical abnormalities during the last 2 years preceding referral were enrolled in the study. At the time of enrollment, information related to demographic, reproductive, and selected behavior factors were collected. From the original cohort of 1128 women, 1023 who had all examinations completed were available for this study. Examination of patients. At the clinic visit after referral a Papanicolaou smear with a Cytobrush endocervical brush (Medscand) and spatula was obtained, and the dacron swab provided with the HPV profile kit was used for collecting an endocervical sample for HPV testing. This was also used to provide a sample for PCR. In addition, a cervical vaginal lavage with 10 ml of normal saline solution was obtained. Colposcopy examination indicated any abnormal areas for cervical biopsies, which were taken after an endocervical curettage. The colposcopy findings and sites of biopsies were recorded on a standardized form. A blood sample was obtained from women who agreed to blood drawing (not an enrollment condition). Diagnostic tests. The enrollment Papanicolaou smear was classified according to the Bethesda classification system in the same laboratory by a single individual. The same person evaluated all biopsy specimens (R.H.K.), which were reported as HPV changes; cervical intraepithelial neoplasia grade 1, 2, or 3; carcinoma in situ; invasive squamous cell carcinoma; adenodysplasia; adenocarcinoma in situ; or adenocarcinoma. The PCR technique was used for HPV screening. HPV DNA was amplified by a general primer pair within the L1 open-reading frame. The primers MY09/MY 11 were
Material and methods The population studied has been previously described.2 Briefly, patients with a Papanicolaou smear reported as high-grade squamous intraepithelial lesion or two smears with atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesion were referred from city and county clinics caring for the indigent population in Harris County, Texas, to the colposcopy clinic in the Department of Obstetrics and Gynecology of Baylor College of Medicine in Houston, Table I. Type-specific primers Type-specific primers 6.1 6.2 11.1 11.2 16.1 16.2 18.1 18.2 31.1 31.2 33.1 33.2 35.1 35.2
Primer sequence
Reference
+5´ TAGTGGGCCTATGGCTCGTC 3´ –5´ TCCATTAGCCTCCACGGGTG 3´ +5´ GGAATACATGCGCCATGTGG 3´ –5´ CGAGCAGACGTCCGTCCTCG 3´ +5´ TGCTAGTGCTTATGCAGCAA 3´ –5´ ATTTACTGCAACATTGGTAC 3´ +5´ AAGGATGCTGCACCGGCTGA 3´ –5´ CACGCACACGCTTGGCAGGT 3´ +5´ ATGGTGATGTACACAACACC 3´ –5´ GTAGTTGCAGGACAACTGAC 3´ +5´ ATGATAGATGATGTAACGCC 3´ –5´ AGTATTTTCTCCTGCACTGC 3´ +5´ CAGAAGTGGACAGACATTGT 3´ –5´ ATGCATACTCCATATGGCTG 3´
5 5 5 5 5 6 7
Table II. HPV types related to histologic findings HPV types
Histologic finding No CIN CIN 1/HPV CIN 2 CIN 3/ CIS Invasive carcinoma# All χ2 for trend Statistical significance
No. 269 477 134 123 4 1007
16* 66 (24.5) 176 (36.9)ll 67 (50.0) 64 (52.0)¶ 373 (37.0) 36.35 p < 0.001
18† 12 (4.5) 22 (4.6) 4 (3.0) 5 (4.1) 1 (25.0) 44 (4.4)
31, 33, 35‡ 10 (3.7) 22 (4.6) 9 (6.7) 5 (4.1) 46 (4.6)
Any high-risk type
All multiple high-risk types
88 (32.7) 220 (46.1) 80 (59.7) 74 (60.2) 1 (25.0) 463 (46.0) 35.98 p < 0.001
38 (14.1) 103 (21.6)ll 28 (20.9) 26 (21.1) 195 (19.4)
Number in parentheses is percent. OR, Odds ratio; CI, confidence interval; CIN, cervical intraepithelial neoplasia; CIS, carcinoma in situ. *Includes type 16 and its combinations with other types. †Includes type 18 and its combinations with types 31, 33, ‡Includes type 31 or 33 or 35. §Odds ratio for any high-risk type. llIncludes one case of adenodysplasia. ¶Includes one case of adenocarcinoma in situ. #Not included in trend analysis. **Includes one case of adenocarcinoma.
and 35.
Adam et al. 1237
Volume 178, Number 6 Am J Obstet Gynecol
used to identify known HPV types as 6, 11, 16, 18, 30, 31, 33, 35, 39, 40, 42, 43, 45, 51, 52, 53, 54, 55, 57, 58, and 59 and 20 unidentified types.3, 4 All samples with negative results with the consensus primers were tested for β-globin with primers GH20 and PC04 for amplification.4 The samples positive for HPV DNA were then assayed with type-specific primers for types 16, 18, 31, 33, and 35. Samples negative for these high-risk types were then tested for low-risk types with primers for types 6 and 11 (Table I).5-7 The PCR conditions used for consensus primers were preheating to 94° C for 5 minutes, and a total of 35 PCR amplification cycles were performed with use of thermocycle-step parameters of 94° C for 30 seconds, 55° C for 30 seconds, and 72° C for 1 minute. The conditions used for the type-specific primers were the same except that 40 amplification cycles were applied. The amplification products were electrophoresed in a 3% agarose gel and were visualized by ethidium bromide. Both HPV testing protocols were performed blinded. Statistical methods. The differences in proportions were calculated by Yates’ corrected χ2 statistics or twotailed Fisher’s exact tests when indicated. Cornfield’s confidence limits were obtained for the odds ratio or exact limits if indicated.8 Positive predictive value, sensitivity, and specificity of the HPV test were calculated for cervical intraepithelial neoplasia grades 2 or 3 in the biopsy specimen.9 Results Three samples were not available for PCR testing, and 13 samples were negative for β-globin. Thus 1007 women were included in this study.
The distribution of high-risk types of HPV detected by PCR was similar in all races. The relationship of different HPV types identified by PCR to histologic findings on enrollment is illustrated in Table II. HPV type 16 alone or in combination with other high-risk types was found in 373 (37.0%) of all women examined or in 56.9% of 655 women with any HPV DNA present in the specimen tested. HPV 16 was identified in 24.5% of women without significant biopsy findings, gradually increasing by severity of biopsy grade to 52.0% in women with cervical intraepithelial neoplasia grade 2 or 3. This trend was highly significant (p < 0.001). HPV 18 was detected in 4.4% and types 31, 33, and 35 in 4.6% of women, respectively, representing 13.7% of all HPVpositive cases. There were 463 (46.0%) samples with at least one high-risk type identified. A total of 70.7% of all samples were positive for HPV by consensus primers. Of these 463 high-risk HPV-positive women, 80.6% had HPV 16. Multiple high-risk types were detected in 19.4% of all women and in 29.8% of samples positive for HPV. Only low-grade HPV types were found in 11.3% of all samples or in 17.4% of the 655 HPV positives. Neither the low-risk nor the unidentified types showed a significant trend with histologic grade. Overall, the odds ratio for any high-risk type gradually increased to 3.11 for cervical intraepithelial neoplasia grade 2 or 3 with a significant trend (p < 0.001). There were 454 women referred to the clinic with smears reported as atypical squamous cells of undetermined significance (n = 35) or low-grade squamous intraepithelial lesions (n = 419) (Table III.) The clinic (follow-up) Papanicolaou smear was reported as negative in
HPV types
6 or 11 27 (10.0) 60 (12.6) 8 (6.0) 18 (14.6)¶ 1 (25.0) 114 (11.3)
Unidentified type 25 (9.3) 33 (6.9) 7 (5.2) 12 (9.8) 1 (25.0) 78 (7.7)
Negative consensus primer 129 (48.0) 164 (33.4) 39 (29.1) 19 (15.4) 1 (25.0)** 352 (35.0) 41.06 p < 0.001
OR§
95% CI
Statistical significance
1.0 1.76 3.05 3.11 0.69
1.27-2.24 1.94-4.79 1.95-4.95 0.01-8.69
p < 0.001 p < 0.001 p < 0.001 p = 1.0
1238 Adam et al.
June 1998 Am J Obstet Gynecol
Table III. Follow-up Papanicolaou smear, high-risk HPV by PCR, and biopsy in 454 women with atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions in referral Papanicolaou smear (N = 454) Follow-up Papanicolaou smear No SIL (n = 266, 58.6%) High-risk HPV–
High-risk HPV+
Biopsy 184 (69.2)* 82 (30.8) No CIN 75 (40.8)† 26 (31.7) CIN 1/HPV 100 (54.3) 40 (48.8)‡ CIN 2/3 9 (4.9)§ 16 (19.5)§ OR for CIN 2/3 4.71 95% CI 1.85-12.22 Statistical significance p < 0.001 Positive predictive value 19.5 Sensitivity 64.0 Specificity 72.6
ASCUS/LSIL (n = 154, 33.9%) High-risk HPV–
HSIL (n = 34, 7.5%)
High-risk HPV+
High-risk HPV–
High-risk HPV+
91 (59.1) 16 (17.6) 60 (65.9) 15 (16.5) 1.18 0.45-3.19 p = 0.88 16.5 62.5 41.5
9 (26.5) 0 4 (44.4) 5 (55.6)
25 (73.5) 2 (8.0) 11 (44.0) 12 (48.0)ll
63 (40.9) 18 (28.6) 36 (57.1) 9 (14.3)
1.00 0.12-4.41 p = 0.74 48.0 70.6 23.5
All (N = 454, 100.0%) High-risk HPV–
High-risk HPV+
256 (56.4) 198 (43.6) 93 (36.3) 44 (22.2) 140 (54.7) 111 (56.1) 23 (9.0) 43 (21.7) 2.81 1.58-5.03 p < 0.001 21.7 65.2 60.1
SIL, Squamous intraepithelial lesions; ASCUS, atypical squamous cells of undetermined significance; LSIL, low-grade squamous intraepithelial lesions; HSIL, high-grade squamous intraepithelial lesions; CIN, carcinoma in situ; OR, odds ratio; CI, confidence interval. *Number in parentheses is percent of column. †Number in parentheses is percent of Papanicolaou diagnoses. ‡Includes one case of adenodysplasia. §Includes one case of adenocarcinoma in situ. llIncludes one case of follow-up Papanicolaou smear showing invasive carcinoma.
58.6% of cases, remained unchanged in 33.9%, and showed a high-grade squamous intraepithelial lesion in 7.5% of women. Of the 266 women who had a negative clinic smear, 25 (9.4%) had biopsy-proved cervical intraepithelial neoplasia grade 2 or 3. In this group, 19.5% of women with high-risk HPV were found to have grade 2 or 3 on biopsy in contrast to only 4.9% of women without high-risk HPV. In 154 women with atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions in the clinic smear, the finding of highgrade cervical intraepithelial neoplasia in biopsy specimen was not significantly different in the high-risk HPV positive and negative patients (14.3% and 16.5%). The same was true for 34 women whose smear was reported as high-grade lesions (55.6% to 48%). A significantly higher odds ratio for cervical intraepithelial neoplasia grade 2 or 3 (4.71%) was observed in women with a negative clinic smear with a positive predictive value of 19.5%, a sensitivity of 64%, and a specificity of 72.6%. No such significant finding can be seen in women with atypical squamous cells of undetermined significance or lowor high-grade squamous intraepithelial lesions in the clinic smear. The sensitivity of the test exceeded 60%, but the specificity dropped substantially to 41.5% and 23.5% in the latter two groups. Five hundred fifty-three women were referred with a Papanicolaou smear reported as high-grade squamous intraepithelial lesions (Table IV). In this group of women
the clinic cytologic findings were reported as normal in 41.0%, atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions in 32.5%, and high-grade squamous intraepithelial lesions in 26.4%. In the 227 women found to have normal clinic cytologic findings, cervical intraepithelial neoplasia grade 2 or 3 was found in a significantly higher proportion (33.7%) in women with high-risk HPV than in women without high-risk HPV (13.2%). There was no such significant difference observed in the other two groups. The most disputed triage approach is in women with atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions reported in their Papanicolaou smears. We compared the results of PCR testing with findings of our previous study with use of the HPV profile assay in this group of women (Table V).2 The HPV profile detected high-risk HPV in 162 (35.7%) specimens, and the PCR was positive for highrisk HPV in 198 women (43.4%). In addition, the PCR detected HPV with consensus primers in 31 women (7%) not identified by the limited number of type-specific primers used in PCR in this study. Both tests had a similar positive predictive value (22.8% and 21.7%), but the PCR had a somewhat higher sensitivity (65.2% compared with 56.1%) but lower specificity (60.1% compared with 67.8%) than the HPV profile in patients found to have cervical intraepithelial neoplasia grade 2 or 3 on the biopsy specimen. The Papanicolaou smear taken at the
Adam et al. 1239
Volume 178, Number 6 Am J Obstet Gynecol
Table IV. Follow-up Papanicolaou smear, high-risk HPV by PCR, and biopsy in 553 women with high-grade squamous intraepithelial lesions in referral Papanicolaou smear (N = 553) Follow-up Papanicolaou smear No SIL (n = 227, 41.0%) High-risk HPV–
ASCUS/LSIL (n = 180, 32.5%)
High-risk HPV+
High-risk HPV–
Biopsy 141 (62.1)* 86 (37.9) No CIN 63 (44.7)† 19 (22.1) CIN 1/HPV 59 (41.8) 38 (44.2) CIN 2/3 19 (13.5)† 29 (33.7) OR for CIN 2/3 3.27 95% CI 1.61-6.66 Statistical significance p < 0.001 Positive predictive value 33.7 Sensitivity 60.4 Specificity 68.2
89 (49.4) 13 (14.6) 44 (49.4) 32 (36.0)
HSIL (n = 146, 26.4%)
High-risk HPV+
91 (50.6) 11 (12.1) 54 (59.3) 26 (28.6) 0.71 0.36-1.40 p = 0.37 28.6 44.8 46.7
High-risk HPV–
High-risk HPV+
58 (39.7) 88 (60.3) 12 (20.7) 14 (15.9) 14 (24.1) 17 (19.3) 32 (55.2) 57 (64.8) 1.49 0.72-3.11 p = 0.32 64.8 64.0 45.6
All (N = 553, 100.0%) High-risk HPV–
High-risk HPV+
288 (52.1) 265 (47.9) 88 (30.6) 44 (16.6) 117 (40.6) 109 (41.1) 83 (28.8) 112 (42.3) 1.81 1.25-2.61 p < 0.005 42.3 57.4 57.3
SIL, Squamous intraepithelial lesions; ASCUS, atypical squamous cells of undetermined significance; HSIL, high-grade squamous intraepithelial lesions; CIN, cervical intraepithelial neoplasia; OR, odds ratio; CI, confidence interval. *Number in parentheses is percent of column. †Number in parentheses is percent of Papanicolaou diagnoses. ‡Includes one case of adenocarcinoma.
Table V. Histologic findings and high-risk HPV in 454 women referred with atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions on cytologic study Histologic findings High-risk HPV
No.
No CIN (n = 137, 30.2%)
HPV profile
Positive Negative
162 292
39 (24.1) 98 (33.6)
86 (53.1) 165 (56.5)*
37 (22.8) 29 (9.9)†
PCR
Positive Negative
198 256
44 (22.2) 93 (36.3)
111 (56.1)* 140 (54.7)
43 (21.7)‡ 23 (9.0)‡
Cytologic findings (clinic Papanicolaou smear)
ASCUS or worse Negative
188
36 (19.1)
111 (59.0)
41 (21.8)
266
101 (38.0)
140 (52.6)*
25 (9.4)†
HPV assay
CIN 1/HPV (n = 251, 55.3%)
CIN 2/3 (n = 66, 14.5%)
OR for CIN 2/3, 95% CI, statistical significance 2.68, 1.53-4.72, p < 0.001 2.81, 1.58-5.03, p < 0.001 2.69, 1.52-4.77, p < 0.001
Screening properties PPV 22.8, SE 56.1, SP 67.8 PPV 21.7, SE, 65.2, SP 60.1 PPV 21.8, SE 62.1, SP 62.1
Number in parentheses is percent of row. CIN, Cervical intraepithelial neoplasia; OR, odds ratio; CI, confidence interval; PPV, positive predictive value; SE, sensitivity; SP, specificity; ASCUS, atypical squamous cells of undetermined significance. *Includes one case of adenodysplasia. †Includes one case of adenocarcinoma in situ. ‡Includes two cases of adenocarcinoma in situ.
time of colposcopy had very similar screening properties with a positive predictive value of 21.8%, a sensitivity of 62.1%, and a specificity of 62.1%. Comment In our previous study of the same population2 we concluded that the approved commercial HPV profile test kit is not sensitive enough for the routine screening of women with atypical Papanicolaou smears for the purpose of triaging patients for immediate colposcopy. Therefore we used the PCR assay, considered the most
sensitive method for HPV DNA detection. The most frequent types of high or intermediate risk (types 16, 18, 31, 33, and 35) and the two most common low-risk HPV types (types 6 and 11) were studied. Our main interest was in finding the prevalence of high-risk types; therefore testing for low-risk types was done only in samples negative for high-risk types. Thus multiple types, when detected, do not include types 6 and 11. When we associate the HPV findings with biopsy findings, it is apparent that 57% of all HPV-positive samples were represented by type 16 and an additional 13.7%
1240 Adam et al.
were of other high-risk types. HPV 16 was identified in 80.6% of patients who had any high-risk types detected. The increasing trend of prevalence of type 16 by higher grades of severity of cervical intraepithelial neoplasia was significant. Women referred to the colposcopy clinic with atypical squamous cells of undetermined significance or lowgrade squamous intraepithelial lesions were seen by different physicians, and the referral cytology examinations were performed in different laboratories. The clinic Papanicolaou smears and biopsy specimens taken at the time of colposcopy were processed in the same laboratory and evaluated by the same pathologist (R.H.K.) without knowing the HPV status of the patient. Thus, if there were any bias in the diagnosis, it would occur with consistency. Women referred with a smear reported as atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions had a negative clinic Papanicolaou smear in 58.6% of cases. It is possible that the referral Papanicolaou smear was in many cases overdiagnosed; on the other hand, the clinic smear might have had, in some cases, false-negative results because of sampling differences. Nevertheless, in this group of patients the incidence of biopsy-proved cervical intraepithelial neoplasia grades 2 and 3 was relatively low—9.4% compared with that in women whose repeat Papanicolaou smears were reported as atypical squamous cells of undetermined significance or low- (15.6%) or high-grade squamous intraepithelial lesions (50%). Similar findings were observed in women referred for high-grade squamous intraepithelial lesions in their Papanicolaou smears. The clinic smear was within the normal range in 41% of all the patients studied. In 21.1% of these women cervical intraepithelial neoplasia grade 2 or 3 was found in the biopsy specimen. Women with atypical squamous cells of undetermined significance or lowgrade squamous intraepithelial lesions in the clinic Papanicolaou smear had cervical intraepithelial neoplasia grade 2 or 3 in 32.2% of their biopsy specimens. A total of 61% of women whose clinic smear was reported as high-grade lesions were found to have cervical intraepithelial neoplasia grade 2 or 3 on the biopsy specimen. The question arises as to how HPV testing would contribute to the detection of cervical intraepithelial neoplasia grades 2 or 3. Women with normal cytologic findings in the clinic smear, whether they were referred with lowor high-grade squamous intraepithelial lesions and detectable high-risk HPV DNA, had a significantly higher rate of biopsy-proved grades 2 and 3 neoplasia than did women without HPV infection. In contrast, women with any cytologic abnormality in the clinic smear had a frequency of cervical intraepithelial neoplasia grade 2 or 3 in the biopsy specimen irrespective of their HPV status. Thus the HPV testing appears helpful for the proper diagnosis and treatment in women with a normal follow-up
June 1998 Am J Obstet Gynecol
Papanicolaou smear after a previous abnormal smear was found. In spite of this, a significant number of patients with a negative HPV test and a negative follow-up smear were found to have grade 2 or 3 neoplasia on the biopsy specimen: 9 of 25 (36%) women referred with atypical squamous cells of undetermined significance or lowgrade squamous intraepithelial lesions and 19 of 48 (39.6%) referred with high-grade squamous intraepithelial lesions. Our results indicate that a repeat Papanicolaou smear showing a similar or more severe abnormality than the first one, irrespective of the HPV testing, is an indication for colposcopy with directed biopsies. A clinic smear at the time of colposcopy does not appear to contribute to further management, as previously noted by Young et al.10 Evidently, the sensitivity and specificity of the HPV assay are of great importance. The comparison of the two HPV assays used in our study also shows a difference, with the PCR being more sensitive in general. The PCR in our study detected HPV in 154 of 257 patients with cervical intraepithelial neoplasia grade 2 or 3. In addition, 19 women (7.3%) were found to have unidentified HPV types. Cox et al.,11 on using the Hybrid Capture test, reported 14 (93.3%) high-risk HPV positives in 15 women with grade 2 or 3 neoplasia on the biopsy specimen. Schneider et al.,12 with use of the same assay, found only 50% (19/38) high-risk HPV in biopsy-proved grade 2 or 3 neoplasia. Wright et al.,13 also using the Hybrid Capture test, found high-risk HPV in 6 of 11 (55%) biopsy-proved grade 2 or 3 neoplasia in women referred because of a smear reported as atypical squamous cells of undetermined significance and in 33 of 39 (84.6%) with grade 2 or 3 neoplasia on the biopsy specimen in women referred for low-grade squamous intraepithelial lesions. Sun et al.14 reported high-risk HPV in 79% of 105 cases with grade 2 or 3 neoplasia or cancer. With use of a PCR assay Cuzick et al.15 identified high-risk HPV types in 58% of women with grade 2 neoplasia and in 84% of women with grade 3 neoplasia on histologic examination. There is a considerable variation in the sensitivities for HPV testing even when a single test is used, possibly because of patient selection or study design. There is no confirmed evidence yet that HPV causes all invasive cervical cancers, although such a possibility cannot be excluded.16 Recently Shah et al.17 did not find differences in the presence of high-risk HPV in patients with cervical intraepithelial neoplasia grade 3 on the biopsy specimen that did or did not progress to invasive cancer. Kiviat and Koutsky18 and Koutsky et al.19 concluded that cervical intraepithelial neoplasia is an early manifestation of HPV infection. They also found, after adjustment for HPV, an association of chlamydia and cytomegalovirus infection with development of cervical neoplasia. Downey et al.20 reported that development of cervical intraepithelial neoplasia grade 2 or 3 is higher in
Volume 178, Number 6 Am J Obstet Gynecol
women with biopsy-proved grade 1 neoplasia at study entry than in women with normal histologic features. There was no difference in disease-free probability in relation to HPV-16 infection. Wheeler et al.21 reported on the limitation of a single-point detection of HPV DNA for assessment of a woman’s HPV status. All these observations definitely indicate a major but not exclusive role of HPV infection in cervical neoplasia. Assays identifying a higher proportion of high-risk HPV in cervical intraepithelial neoplasia grade 2 or 3 also detect an increased prevalence of this infection in grade 1 neoplasia and even in women without significant pathologic findings. To test all patients with abnormal Papanicolaou smears for high-risk HPV has its advantages and substantial drawbacks. As mentioned, a positive test would support the necessity of a colposcopic examination and enhance the detection of a grade 2 or 3 neoplasia on the biopsy specimen of a woman referred for a cytologic abnormality who has a follow-up smear reported as normal. On the other hand, solely relying on these tests would still miss a considerable number of patients with grade 2 or 3 neoplasia. So far, there is no good evidence to support the cost effectiveness of HPV testing in clinical practice.2 The majority of cases of cervical intraepithelial neoplasia grade 1 regress, including many cases in our experience with detected high-risk HPV. A positive test indicates a higher probability of a diagnosis of grade 2 or 3 neoplasia, but a negative test does not exclude it. Thus the use of these tests at present should remain in the research domain.
REFERENCES
1. Zur Hausen H. Human papillomaviruses in the pathogenesis of anogenital cancer. Virology 1991;184:9-13. 2. Kaufman RH, Adam E, Icenogle J. Relevance of human papillomavirus screening in management of cervical intraepithelial neoplasia. Am J Obstet Gynecol 1997;176:87-92. 3. Manos MM, Ting Y, Wright DK, Lewis AJ, Broker TR, Wolinsky SM. Cancer cells 7: molecular diagnostics of human cancer. Cold Spring Harbor: Cold Spring Harbor Laboratory; 1989. 4. Bauer HM, Ting Y, Greer CE, Chambers JC, Tashiro CJ, Chimera J, et al. Genital human papillomavirus infection in female university students as determined by a PCR-based method. JAMA 1991;265:472-7. 5. Van der Brule AJC, Meijer CLM, Bakels V, Kencmans P, Walboomers JMM. Rapid selection of human papillomavirus in cervical scrapes by combined general primer-mediated and type-specific polymerase chain reaction. J Clin Microbiol 1990;28:2739-43. 6. Rymark P, Forslund O, Hansson BG, Lindholm K. Genital HPV infection not a local but a regional infection: experience from a female teenage group. J Clin Microbiol 1990;28:739-43. 7. Forslund O, Hansson BG, Rymark P, Bjerrf B. Human papillomavirus DNA in urine samples compared with that in simultaneously collected urethra and cervix samples. Clin Microbiol 1993;31:1975-9. 8. Fless JL. Statistical methods for rates and proportions. New York: John Wiley; 1981. 9. Brookmyer R, Gail MH. AIDS epidemiology: a quantitative approach. New York: Oxford University Press; 1994.
Adam et al. 1241
10. Young NA, Naryshkin S, Bowman RL. Value of repeat cervical smear at the time of colposcopic biopsy. Diagn Cytopathol 1993;9:646-9. 11. Cox JT, Lorincz AT, Schiffman MH. Human papillomavirus testing by hybrid capture appears to be useful in triaging women with a cytologic diagnosis of atypical squamous cells of undetermined significance. Am J Obstet Gynecol 1995;172:946-54. 12. Schneider A, Zahn DM, Kirchmayr R. Screening for intraepithelial neoplasia grade 2/3: validity of cytologic study, cervicography, and human papilloma virus detection. Am J Obstet Gynecol 1996;174:1534-41. 13. Wright TC, Sun XW, Koulos J. Comparison of management algorithms for the evaluation of women with low-grade cytologic abnormalities. Obstet Gynecol 1995;85:202-10. 14. Sun X-W, Ferenczy A, Johnson D, Koulos JP, Lungu O, Richart RM, et al. Evaluation of the hybrid capture human papilloma virus deoxyribonucleic acid detection test. Am J Obstet Gynecol 1995;173:1432-37. 15. Cuzick J, Terry G, Ho L, Hollingworth T, Anderson M. Type-specific human papillomavirus DNA in abnormal smears as a predictor of high-grade cervical intraepithelial neoplasia. Br J Cancer 1994;69:167-71. 16. Zur Hausen H. Are human papillomavirus infections not necessary or sufficient causal factors for invasive cancer of the cervix? Int J Cancer 1995;63:315-6. 17. Shah KV, Kessis TD, Shah F, Gupta JW, Shibata D, Jones RW. Human papillomavirus investigation of patients with cervical intraepithelial neoplasia 3, some of whom progressed to invasive cancer. Int J Gynecol Pathol 1996;15:127-30. 18. Kiviat NB, Koutsky LA. Specific human papillomavirus types as the causal agents of most cervical intraepithelial neoplasia: implications for current views and treatment. J Natl Cancer Inst 1993;85:934-5. 19. Koutsky LA, Holmes KK, Critchlow CW, Stevens CE, Paavonen S, Beckman AM. A cohort study of the risk of cervical intraepithelial neoplasia grade II or III in relation to papillomavirus. N Engl J Med 1992;327:1272-8. 20. Downey GP, Bavin PJ, Deery ARS, Crow J, Griffiths PD, Emery VC, et al. Relation between human papillomavirus type 16 and potential for progression of minor-grade cervical disease. Lancet 1994;344:432-5. 21. Wheeler CM, Greer CE, Becker TM, Hunt WC, Anderson SM, Manos MM. Short-term fluctuations in the detection of cervical human papillomavirus DNA. Obstet Gynecol 1996;88:261-8.
Discussion DR. MICHAEL HOPKINS, Akron, Ohio. I believe this report will influence our attitude toward the connection of HPV to premalignant disease of the cervix. The authors tried to determine whether HPV screening would help identify high-risk groups and their article contains a wealth of data with tables that extensively detail their results. The methods have been described and are appropriate for their population. There is no long-term patient follow-up. Thus their data are valid only for the incidence of cervical intraepithelial neoplasia and HPV at the initial colposcopic visit. In their report the authors compared referral Papanicolaou smear with follow-up Papanicolaou smear, both smears with biopsy specimen results; both smears with HPV results, and, last, biopsy specimen results with HPV results. In their analysis of >1000 patients a number of interesting details emerged and along with this a number of questions. The authors’ evaluated referred patients with colposcopic-directed biopsies. A PCR analysis was then performed for HPV. Interestingly, among patients with no dysplasia one third had high-risk HPV types. Also, one
1242 Adam et al.
half these patients without dysplasia did not have any identifiable HPV. Among the patients with moderate-severe dysplasia or carcinoma in situ, 60% had high-risk HPV. Interestingly, 25% had no identifiable HPV. Thus it would appear that identifying high-risk subtypes does not predict high-grade squamous intraepithelial lesions. Another interesting finding was the results when a follow-up or new Papanicolaou smear was done at the visit for colposcopy. When comparing the referral smear and the follow-up smear, a large discrepancy exists between the referral smear and the in-house or follow-up cytologic findings. Almost 500 patients each are in the lowgrade or high-grade referral groups. In the low-grade referral group, 60% had negative cytologic findings on repeat Papanicolaou smear, and only one third of the time did the low-grade cytologic findings agree with the referral cytologic findings. In >500 patients with highgrade changes 40% had negative cytologic findings and 33% had low-grade changes. Thus only 25% were confirmed to have high-grade changes on cytologic study, whereas 75%, an exceedingly large number, had a difference in their cytologic findings. Such widespread variations certainly highlight the inexact science of cytology. This, combined with the broad distribution of highrisk HPV types across the groups, is good evidence that colposcopy with biopsy will continue to be the standard for evaluation of abnormal cytologic findings. I took the liberty of combining some tables from the authors’ article and present these to highlight what I believe is a continuing problem in our management of premalignant disease. When using the new follow-up cytologic results, the authors report a very high rate of negative findings, although approximately 15% of these patients with negative results have moderate or severe dysplasia on the biopsy specimen. Among those with atypical squamous cells of undetermined significance or low-grade changes, approximately 25% have high-grade histologic findings. This is a continuing management problem with cytology. Most authors report an incidence in the 10% to 20% range of significant premalignant finding with low-grade cytologic findings. This highlights the problems with clinical management: The Papanicolaou smear is a very inexact science at best; a third of patients with no cervical intraepithelial neoplasia findings have high-risk HPV; patients may be exposed to a new viral agent that is unknown to the examining physician. I ask the following questions: (1) Was there any prolonged delay in the time from the referral Papanicolaou smear to the Papanicolaou smear obtained at colposcopy in the clinic? Also, were any of the referral smears done in the antepartum period with the patient not being seen in colposcopy clinic until the postpartum period? (2) Why is there such a large difference between the referral cytologic findings and those obtained at colposcopy? (3) Do the authors continue to do Papanicolaou smears at colposcopy? (4) How can we explain such a high rate of high-risk HPV in patients without any cervical intraepithelial neoplasia and what may be considered a low rate of high-risk HPV (grade 2 or 3) in patients with cervical intraepithelial neoplasia grade 3 or carcinoma in situ?
June 1998 Am J Obstet Gynecol
In closing, this is a significant report that suggests that perhaps HPV is not the complete answer to the etiology of premalignant disease. DR. JOHN CALKINS, Kansas City, Kansas. The authors raise an interesting question in their article. Specifically, what role does HPV testing play in triage of patients for identification of significant intraepithelial lesions of the cervix that would warrant treatment? As noted by the authors, we have witnessed a significant decrease in the incidence of cervical cancer since the introduction of Papanicolaou smear screening in the 1950s. It is intuitive that identification of precancerous changes and treatment thereof has played a major role in that reduction. In spite of that impact, the Papanicolaou smear has been much maligned of late. Clinicians are being inundated by changes in the technology, such as Pap-Net and Thin-Prep. Recognizing that HPV is the most plausible pathogenic agent for cervical carcinogenesis, identification of high-risk types is an appealing alternative to aid in selection of patients for more significant evaluation. In the current article the authors report on their experience using PCR in a study population on whom they had previously reported the use of the HPV profile testing kit for aid in triaging patients for immediate colposcopy. In that study they concluded that the HPV profile testing kit was not sensitive enough for routine screening for purposes of immediate triage. Recognizing that PCR is a more sensitive method for HPV detection, they reassessed their study population with use of consensus primers meant to identify in excess of 40 subtypes of HPV from samples obtained during the screening phase of their study. They did identify a significantly higher odds ratio for high-grade intraepithelial neoplasia in women with atypical squamous cells of undetermined significance and low-grade squamous intraepithelial lesions on referral Papanicolaou smears if their clinic smears were negative and their HPV testing was positive (with a positive predictive value of 19.5%, a sensitivity of 64%, and a specificity of 72.6%). However, if their clinic Papanicolaou smear was abnormal, then no such significant finding could be identified. When PCR was contrasted with the HPV profile detection kit, PCR was indeed positive in a higher percentage of women, 43% versus 36%. Both tests had similar positive predictor values, but the PCR was found to be more sensitive but less specific. The authors concluded that a positive HPV test indicates a higher probability of highgrade cervical intraepithelial neoplasia but does not exclude such a condition, and thus the test still should be viewed as a research tool. I will focus my questions on three areas. The first area relates to patient selection and specifically the referral Papanicolaou smear. Since the advent of the Bethesda system, many of us who provide colposcopic services have been plagued by an increase in referrals, largely as a result of the designation atypical squamous cells of undetermined significance. To triage patients for colposcopy, we will commonly repeat cytologic studies for this finding and for low-grade squamous intraepithelial lesions and
Adam et al. 1243
Volume 178, Number 6 Am J Obstet Gynecol
further evaluate only those patients with persistent squamous intraepithelial changes. As an individual who works directly with county health departments in the Kansas City area, I can attest to their anxiety about “observation” of minor Papanicolaou smear abnormalities, especially because of patient compliance. Looking at the portion of the patients in this study with these findings, 15% of the patients were found to have a high-grade lesion, more than one third of which would have been missed by PCR screening. The referral criteria in the current study could be considered relatively conservative because they required either a high-grade Papanicolaou smear result or consecutive findings of atypical squamous cells of undetermined significance or low-grade lesions for study entry. Interestingly, in the study design all patients received a “clinic Papanicolaou smear” and in that group almost half were read as normal. Do the authors have an idea regarding the incidence of atypical squamous cells of undetermined significance and low-grade squamous intraepithelial lesions in their county clinic study population, especially in light of the high incidence of negative Papanicolaou smear findings during the screening phase of their study? Speaking from my own experience, I had developed what I thought was a reasonable triage mechanism on the basis of the incidence of these in my own population. Unfortunately, with the advent of managed care, we are now required to use multiple laboratories for cytologic testing, which makes it much more difficult to track percentages and consequently continue to use such triage mechanisms. My second question relates to study design and specifically “collection method.” The authors identified highrisk HPV types in roughly 60% of patients with highgrade disease on the biopsy specimen. Hatch et al.1 in a similar study published in 1995 using Hybrid Capture technology found a 74% incidence of high-risk HPV types with high-grade squamous intraepithelial lesions. Nuovo et al.2 a few years earlier quoted in excess of 90% positivity with high-grade lesions. Do the authors feel that the study design, specifically obtaining a Papanicolaou smear before HPV swab, may in any way affect the positivity of their detection rate? Additionally, do the authors feel that the timing of HPV collection is of significance? Wheeler et al.3 in 1996 published a report on short-term fluctuations in the detection of cervical HPV DNA with use of various methods. Specifically, with PCR they found a range between 21% and 47% monthto-month in their study population of 72 college-aged women, roughly one third of which had a history of an abnormal Papanicolaou smear. My third question relates to type-specific data identified in the article The authors identified type 16 in 57% of all HPV-positive samples and in excess of 80% of patients who had any high-risk type of HPV DNA detected. They noted an increasing trend of prevalence of type 16 by higher grades of severity of cervical intraepithelial neoplasia and concluded that it was significant. However, they felt that no such trend was observed with other HPV types. In light of the fact that the other high-risk HPV
types were identified alone in only single digit numbers of patients, I would question whether there is sufficient power to draw such conclusions. I would ask the authors if a power analysis was performed with regard to those conclusions. In conclusion, I applaud the authors in their attempt to improve the screening approach to patients with intraepithelial disease of the cervix and look forward to their follow-up data as it relates to screening protocols. Their conclusions support my personal opinion, that at this time HPV testing is a technique still in search of useful clinical application. REFERENCES
1. Hatch KD, Schneider A, Abdel-Nour MW. An evaluation of human papillomavirus testing for intermediate and high-risk types as triage before colposcopy. Am J Obstet Gynecol 1996;172:1150-7. 2. Nuovo GJ, Blanco JS, Leipzig S, Smith D. Human papillomavirus detection in cervical lesions nondiagnostic for cervical intraepithelial neoplasia: correlation with Papanicolaou smear, colposcopy, and occurrence of cervical intraepithelial neoplasia. Obstet Gynecol 1990;75:1006-11. 3. Wheeler CM, Greer CE, Becker TM, et al. Short-term fluctuations in the detection of cervical human papillomavirus DNA. Obstet Gynecol 1996;88:261.
DR. ROBERT J. CARPENTER, Houston, Texas. Would you or can you address the specific issue of those patients who on repeat Papanicolaou smears had negative results and what you think may have occurred in those that had been submitted to one of the forms of automated Papanicolaou smear review, whether you would have then seen a similar distribution to your biopsy data? DR. MELVIN V. GERBIE, Chicago, Illinois. How do you at this point counsel those 32% percent of patients who had negative biopsy specimens and yet still showed highrisk HPV changes? DR. KAUFMAN (Closing). The first comment was related to the time interval between the smears and how many were done ante partum or post partum. The time interval varied between 2 months and 1 year. The average was probably about 4 months between the time the last Papanicolaou smear was taken and the patients were seen in our clinic. There were very few of this population of individuals who were antepartum. I cannot give you the exact number, but I would say it’s probably <2% of the entire group. Why the difference between referral cytologic findings and the findings obtained in our clinic? There were several reasons. One could be a higher sensitivity, if you will, on the part of the pathologists working in the state laboratory. In general, the reporting that comes out of our state laboratory in Austin is extremely good. In fact, I did go back and review the majority of these referral smears; in many cases or in most cases there was general agreement, but in about 40% to 50% I tended to downgrade the reports. That may refer to the fact that I am undercalling or it could be the fact that they are overcalling, and that’s a question that I don’t think I can answer.
1244 Adam et al.
Also, another related factor is the technique of taking the smears. The smears were taken in our clinic by residents with faculty supervision, and the smears that went to the state laboratory were taken by a whole variety of individuals, family practitioners, obstetricians, nurse practitioners, etc. Do we still do Papanicolaou smears at the time of the colposcopy? In our clinic we do. There have been several studies to suggest that really is a waste of time. If you have a significantly abnormal smear and the patient is brought in for colposcopy, why bother to do another smear? It doesn’t really in the long run contribute to how you’re going to manage the patient. On the other hand, a point could be raised that if the colposcopic examination is normal, you don’t see anything abnormal, and you obtain tissue that does not show pathologic features, and the repeat Papanicolaou smear suggests high-grade disease, you better keep looking; whereas if it is reported as negative and your examination is negative and the biopsy specimens are negative, I think you can relax a bit and probably defer seeing the patient again for at least another 6 months. The question was raised as to high-risk HPV seen in a significant number of our patients with no pathologic features. I think this partly relates to the population of patients we are seeing. We are dealing with a very-highrisk population of individuals, and the prevalence of high-risk HPV in this population is quite high. Overall, in the group that we studied here, about 11% were found to have low-risk HPV; therefore 89% to 90% had a high-risk HPV type. In our population this is what we see. If you look at most studies that are done to examine the prevalence of HPV in a community, it may run as high as 40% or even higher in a given community, especially in a high-risk community such as we’re dealing with. The question was raised as to the prevalence of Papanicolaou smears reported as cervical intraepithelial neoplasia in our population. It again depends on which laboratory is doing the reporting. Somewhere between about 6% and 8% of the patients in our clinic population will have smears reported as atypical squamous cells of undetermined significance or higher. Did the method of collection have anything to do with it? Theoretically, with PCR it should not. It’s such a sensitive technique that even if you do a Papanicolaou smear first, you should expect to find plenty of cells with HPV DNA if, in fact, HPV DNA is present. The other point I think that relates to some of our findings is why we saw such a significant number of individuals with no high-risk HPV and yet grade 2 or 3 cervical intraepithelial neoplasia is the age of our population. About 30+%, a third of the patients, were >30 years old; our data, which have not yet been published, would sug-
June 1998 Am J Obstet Gynecol
gest that there was a distinct, significant decrease in the frequency of detection of HPV as women got older. In other words, in women >30 years old, the frequency of HPV detection was significantly less than it was in the group of patients between ages 15 and 25 years, for example, whereas the frequency of grade 2 and 3 cervical intraepithelial neoplasia increased significantly from the younger group to the older group. Therefore, we have a group of women >30 years old, many of whom were found to have grade 2 or 3 lesions and many of whom had negative HPV testing. We can discuss for another hour why that is. Is the HPV burned out, no longer present? Is it not detectable? I don’t think I can answer that at the present time, but those are just our findings. I don’t think we can draw any distinct conclusions about the other HPV types in relation to finding grade 2 or 3 cervical intraepithelial neoplasia. Certainly, when you lump all the high-risk HPV types, there’s a definite significant trend to find high-risk HPV in higher grades of cervical intraepithelial neoplasia. When it comes to specific types, besides type 16, I don’t think we had enough cases in the 18, 31, 33, and 35 group to come to any conclusions. Dr. Carpenter raised the point about whether automated cytologic studies would have changed some of our results; they may well have. In fact, we just completed a study sending a series of patients who had Papanicolaou smears reported as atypical squamous cells of undetermined significance in our laboratory for review; in 20% of the cases the smears were upgraded. All these patients, I might add, had had biopsies within a year’s time of the Papanicolaou smear being taken. There were 9 patients who were found to have grade 3 lesions, and in 6 of the 9 the smear was upgraded by automated cytologic study. As a matter of fact, right now we are in the midst of a study reviewing smears in our clinic reported as atypical squamous cells of undetermined significance or low-grade squamous intraepithelial lesions. We tried to relate our findings to absence or presence of high-risk HPV. Dr. Gerbie asked about counseling in the patients who have a negative biopsy specimen and high-risk HPV. I just tell them they have this. The biopsy specimens are negative, and we’ll continue to follow them up. If the Papanicolaou smear stays negative, there’s no reason to be concerned, and as long as there’s nothing we can see in the biopsy, the likelihood of transmitting the infection is small. I don’t say it’s impossible because I don’t know; and we will follow them up very carefully at regular intervals, 6 months to 1 year, depending on what we found. If they have ever had an abnormal Papanicolaou smear, it will be 6-month intervals until we have three consecutive negative smears. Otherwise, we will see them at yearly intervals.